Using Richardson Extrapolation to Improve the Accuracy of Processing and Analyzing Empirical Data

Abstract

We investigate properties of empirical data under random uncertainty. A new approach is proposed to improve the accuracy of constructing a probability density function and estimating its error. The method uses Richardson extrapolation and Runge rules for kernel estimates with various smoothing options. It is shown that the application of the Runge rules enables evaluating the error of kernel estimates for a probability density function and the value of its second derivative.